OUTDOOR WEATHERING OF GLASS FIBRE REINFORCED

POLYESTER AND VINYL ESTER COMPOSITES:

FIRST RESULTS AFTER ONE YEAR EXPOSURE

Anu Heikkilä1, Johanna Juhola2, Merja Kaunismaa2, Jehki Hakkarainen3, Outi Meinander1,
Uwe Feister4, Stelios Kazadzis5, Alkis Bais5, José Manuel Vilaplana Guerrero6, Carmen
Guirado7, Juanjo Rodríguez7, Petri Kärhä8, Timo Ture2, Seppo Syrjälä3, Jarkko Koskinen1
1
Finnish Meteorological Institute, R&D / Earth Observation, FI-00101 Helsinki, Finland
2
Elastopoli Oy, FI-38210 Vammala, Finland
3
Tampere University of Technology, Lab. of Plastics and Elastomer Technology, FI-33101 Tampere, Finland
4
Deutscher Wetterdienst, Met. Obs. Lindenberg, Am Observatorium 12, 15848 Lindenberg, Germany
5
Laboratory of Atmospheric Physics, Aristotle University Thessaloniki, 54124, Thessaloniki, Greece
6
Atmospheric Sounding Station "El Arenosillo", San Juan del Puerto Matalacañas Road Km33,
21130 Mazagon, Huelva, Spain
7
Agencia Estatal de Meteorología, Izaña ARC, C/La Marina 20 Planta 6, 38071 Santa Cruz de Tenerife, Spain
8
Helsinki University of Technology, Metrology Research Institute, FI-02015 TKK, Finland

ABSTRACT
In the current paper, first results of a long-term outdoor exposure tests of glass fibre reinforced polyester and
vinyl ester composites, conducted at seven European sites, are reported. Aged specimens from the first three
exposure periods were tested for their colour, gloss, hardness and flexural strength. Relative changes in the
properties tested were examined as functions of accumulated UV exposures. Ageing environments were
ranked according to the severity of the UV exposure and other major environmental stress factors:
temperature, precipitation, humidity and tropospheric ozone concentration. Spearman rank correlation
analysis was applied on the relative property changes in the specimens and the severity of environmental
agents. Statistically significant correlation coefficients were considered indicators for factors mainly
responsible for the observed property change.
Evolution in all properties and especially in hardness and flexural strength was found resin dependent. The
stress factors most severely degrading the materials were found both resin and property specific. The most
significant results in the performed Spearman rank correlation test were obtained for UV exposure causing the
yellowing (in GFR polyester), heat stress contributing both to the yellowing (in GFR polyester and vinyl
ester) and decrease of flexural strength (in GFR polyester), and precipitation also promoting the yellowing (in
GFR vinyl ester). In general, GFR polyester appeared slightly more susceptible to weathering than GFR vinyl
ester over the first year of the experiment. Differences in the performance might be traced to higher resistance
of GFR vinyl ester to two phenomena: Norrish II types photochemical processes and hydrolytic degradation.

1. INTRODUCTION

In spring 2005, a material research project funded by the Finnish Meteorological Institute
and the Finnish Funding Agency for Technology and Innovation was launched. The project
was named UVEMA (UV radiation Effects on MAterials). Within the framework of the
project, a weathering network of seven European sites (Fig. 1) was established. The
network extends from the Canary Islands of Spain (latitude 28.5°N) to the Lapland of
Finland (latitude 67.4°N), covering a wide range of weathering conditions. Since autumn
2005, the sites of the network have been maintaining weathering platforms of specimens of
different kinds of polymeric materials, including glass-fibre-reinforced polyester and vinyl
ester composites. Tests of the weathered samples are to provide unbiased information on
the degradation of the materials in their real outdoor service environments. Environmental
data collected from the test sites over the weathering experiment will also make it possible
to estimate the correlation between accelerated test environments and the natural service
environments.

Figure 1. Outdoor weathering network established within the UVEMA project. Whitish
plates in the racks are the GFR polyester and vinyl ester samples.

In this paper, we will for the first time report on the results of our four-year outdoor
exposure tests of GFR polyester and vinyl ester composites. Results from the first three
epxosure periods of lengths 44-292 days are presented. Weathering characteristics of the
test sites is depicted and changes in material properties as regards to colour, gloss, hardness
and tensile strength are examined. Main stress factors attributable to the observed property
changes are distinguished through the use of Spearman's rank correlation analysis.
Differences between the ageing behaviour of GFR polyester and vinyl ester are discussed.
2. EXPERIMENTAL
2.1 Test sites
The test sites were selected from the sites regularly submitting spectral UV irradiance data
to the European UV database hosted by the Finnish Meteorological Institute
(http://uvdb.fmi.fi/uvdb). In addition to the high quality of the UV measurements, the
selection criteria included regular on-site measurements of the major meteorological
parameters and nearby monitoring of tropospheric ozone concentration. The sites were
selected in such a way that a representative European scale of the weathering parameters
would be covered. The main characteristics of the sites are listed in Table 1.

Table 1. Characteristics of the sites forming the UVEMA network.

Latº Lonº Alt UV-I Level of Relative
Type Site Country
(N) (E) (m) (max) pollution humidity
Sodankylä Finland 67.4 26.6 179 5 low moderate
northern
Jokioinen Finland 60.8 23.5 104 6 low moderate
central Lindenberg Germany 52.2 14.1 100 8 moderate moderate
Ispra Italy 45.8 8.6 214 9 moderate moderate
southern El Arenosillo Spain 37.1 -6.7 41 10 low high
Thessaloniki Greece 40.6 23.0 20 10 high high
extreme Izaña Spain 28.5 -16.5 2367 13 low low

2.2 Exposures
The weathering of the GFR polyester and vinyl ester composites started in autumn 2005.
Until autumn 2007, five specimen batches have been removed from exposure by following
a pre-defined schedule [1] for post-exposure material testing. The schedule was based on
predicted accumulation of UV dose. Start and end dates for the first three exposure periods,
together with the length of the periods, are listed in Table 2.

Table 2. Start dates, end dates and lengths of the first three exposure periods within the
outdoor exosure programme.
Start date Period #1 Period #2 Period #3
Site
(ddmmyy) End date Length (d) End date Length (d) End date Length (d)
Sodankylä 090905 240306 197 070506 241 270606 292
Jokioinen 210906 220306 183 100506 232 020706 285
Lindenberg 270905 130206 140 010506 217 290606 276
Ispra 041005 210106 110 210406 200 270606 267
El Arenosillo 220905 071105 47 100406 201 250606 277
Thessaloniki 170905 221005 36 210306 186 180606 275
Izaña 111005 231105 44 130306 154 190606 252

Measurements of spectral UV irradiance, temperature, humidity, precipitation and

tropospheric ozone concentration were conducted by each test site along the weathering
experiment (Fig. 2). Measured data were collected from the test sites and stored into a
database.
 9 Spectral UV irradiance
9 Temperature
9 Relative humidity
9 Precipitation
9 Ozone concentration
Figure 2. Environmental data collected from the test sites.

2.3 Composite materials

The composites are short glass fibre reinforced polyester and vinyl ester. The materials only
differed from each other in respect of the matrix resin. No stabilizers were used in
manufacturing of the samples.

2.4 Test methods

Surface and bulk property changes of the weathered samples were characterized by
measurements of colour, gloss, hardness and strength. A summary of the test methods is
given in Table 3.

Table 3. Methods used in testing the properties of the exposed samples.

Property tested Equipment Quantity measured
Colour Minolta Chroma Meter CR-200 Colour in Yxy notation
(Y = lightness in %;
xy = colour tones)
Gloss Zehnter Glossmeter 2 GM 75° Gloss angle in degrees
Hardness AFFRI Hardness Tester Hardness in Shore D scale
Strength Messphysik MIDI 10-20/4x11 Flexural strength in MPa

2.5 Data analysis methods

The temporal resolution of the primary environmental data varies from 10 minutes to 3
hours and daily means or sums. For coherent data analysis, suitable secondary quantities
have to be derived from the collected data. For temperature, relative humidity and ozone
concentration, this is performed in two steps. First, daily means are computed from the
primary data series for each site. Next, medians of the daily means are computed for each
exposure period and for each site. To make the figures computed for periods of different
lengths commensurable, the medians are further multiplied with the lengths of the periods.
For precipitation, accumulations over the periods are computed. For UV exposure,
predicted accumulation of CIE erythemally weighted UV dose is used as in exposure
timetable planning. The figures computed and passed to further analysis are summarized in
Table 4.
Table 4. Environmental stress factors and the quantities assumed to represent the stress on
the materials over the exposure periods of varying lengths.
Stress factor Quantity describing the stress over the period Unit
UV radiation Sum of the daily sums kJ m-2
Temperature Median of the daily means × length of the period K days
Precipitation Sum of the daily sums mm
Relative humidity Median of the daily means × length of the period % days
Ozone concentration Median of the daily means × length of the period µg m-3 days

From the test results of the exposed samples, averages of the five measurements are
computed. Averages for gloss, hardness, and flexural strength are used as such in the
following analysis. From the colour measurements, none of the measured values Yxy
appeared suitable for describing the white colour of the unexposed samples turning into
yellowish. Therefore, the yellowness index YI, defined by formula
YI = [(1.28 X − 1.06 Z ) / Y ]× 100 (1)
and earlier used by, e.g., Cho et al. [2], was computed and used as a measure of colour
change. In formula (1), symbols XYZ denote tristimulus values that can be computed from
the Yxy-coordinates by using equations
xY (1 − x − y )Y
X = , Y = Y , and Z = . (2)
y y
Spearman rank correlation analysis is employed to find out the stress factors attributable to
the property changes observed in the exposed samples. The exposure treatments (N=21) are
ranked according to the values of the quantities given in Table 4, i.e., the severity of each
stress factor. Relative property changes, as compared to the unexposed samples, are ranked
similarly. Correlation between the variables (X1,X2) = (rank number of exposure, rank
number of relative property change) is examined by computing the Spearman rank
correlation coefficient ρS. If the null hypothesis is formulated as H0: "X1 and X2 are
independent of each other, ρS = 0", the obvious formulation of the alternative hypotheses is
either H1: "X1 and X2 are positively correlated with each other, ρS > 0" or "X1 and X2 are
negatively correlated with each other, ρS < 0". By performing one-tailed tests of H0 at
confidence levels α = 0.01 and α = 0.05, statistically very significant and significant rank
correlations, respectively, are distinguished.

3. RESULTS AND DISCUSSION

3.1 Exposure environments
Values computed for the quantities listed in Table 4, describing the strength of each
environmental stress factor over each exposure period, were first examined separately from
the data obtained from the tests of the exposed samples. To quantify the average harshness
of the weathering microclimates prevailing by each test site over the exposure periods, the
rank numbers obtained by each site during the three exposure periods were defined. The
resulting numbers were used to arrange the sites from the mildest to the hardest weathering
environments as shown in Table 5.
Table 5. Sites ranked according the severity of the different stresses, number 1(7) in each
column indicating the mildest(hardest) exposure. UV = UV stress; T = heat stress; P =
stress caused by rain; RH = stress caused by moisture; O3 = stress caused by ozone
concentration of the air.
Period #1 Period #2 Period #3
Site
UV T P RH O3 UV T P RH O3 UV T P RH O3
Sodankylä 1 7 7 7 7 1 7 4 6 6 1 6 3 6 4
Jokioinen 2 6 6 6 5 2 6 3 7 4 2 4 2 7 3
Lindenberg 3 5 5 5 6 3 5 5 5 7 3 3 5 5 6
Ispra 4 4 3 4 1 4 3 6 4 1 4 2 6 3 1
El Arenosillo 5 3 2 1 2 5 4 2 2 2 5 7 4 2 2
Thessaloniki 6 2 4 3 3 6 2 7 3 3 6 5 7 4 5
Izaña 7 1 1 2 4 7 1 1 1 5 7 1 1 1 7

The ranks for UV stress is exactly the same for every period, as the timing of the exposure
was adjusted to the accumulation of UV. For the other factors, the ranks differ from period
to period, as expected. The result obtained for heat stresses, especially during the first
period, may appear peculiar at first sight. However, it is explained by the start of exposure
taken place in the autumn and the varying lengths of the periods.

3.2 Property changes in GFRPs

The property changes were first viewed in terms of medians of the relative changes
occurred over the exposure periods. The bar charts are shown in Fig. 3.

Changes in yellowness index Changes in gloss

1200 % 20 %
median of relative changes

median of relative changes

1000 % 15 %

800 %
10 %

600 %
5%
400 %
0%
200 % 1st period 2nd period 3rd period
-5 %
0%
1st period 2nd period 3rd period -10 %

Composite A Composite B Composite A Composite B

Changes in hardness Changes in flexural strength

8% 30 %
median of relative changes
median of relative changes

6%
20 %

4%
10 %
2%
0%
0%
1st period 2nd period 3rd period
1st period 2nd period 3rd period
-10 %
-2 %

-4 % -20 %

Composite A Composite B Composite A Composite B

Figure 3. Medians of the relative property changes of the exposed composite samples over
the first three exposure periods, with Composite A denoting the GFR polyester and
Composite B denoting the GFR vinyl ester.

The change in the yellowness index is remarkable and in the same direction for both GFR
composites. GFR polyester seems to turn yellow more rapidly than GFR vinyl ester.
Changes in gloss include an initial increase of gloss, followed by a decrease as the
weathering proceeds. Again, GFR polyester looses its gloss more rapidly than GFR vinyl
ester. Changes in hardness are not high in percentages, but occur in the opposite directions
for the two types of composites. While the hardness of GFR polyester increases, the
hardness of GFR vinyl ester decreases. Changes in flexural strength repeat the same
feature: GFR vinyl ester gets stronger, but GFR polyester weaker.
The temporal evolution of the properties were next viewed by plotting the observed
changes as functions of predicted accumulation of UV dose. The results are shown in Fig.
4.

Composite B
Composite A

25 35

30
20
sod sod
jok 25 jok
15 lin lin
isp 20 isp
YI
YI

10 the the
ela 15 ela
iza iza
5
10

0 5
0 200 400 600 800 1000 0 200 400 600 800 1000
predicted CIE erythem ally w eighted UV dose (kJ/m 2) predicted CIE erythem ally w eighted UV dose (kJ/m 2)

Composite A Composite B

90 70

85

80 65
sod sod
75 jok jok
gloss angle (deg)
gloss angle (deg)

lin 60 lin
70
isp isp
65
the 55 the
60 ela ela
iza iza
55
50
50

45 45
0 200 400 600 800 1000 0 200 400 600 800 1000
predicted CIE erythem ally w eighted UV dose (kJ/m 2) predicted CIE erythem ally w eighted UV dose (kJ/m 2)
 Composite A Composite B

89 87

86
88
85
sod sod
hardness (Shore D)

hardness (Shore D)
87 jok 84 jok
lin lin
83
86 isp isp
82
the the
85 ela 81 ela
iza iza
80
84
79

83 78
0 200 400 600 800 1000 0 200 400 600 800 1000
predicted CIE erythem ally w eighted UV dose (kJ/m 2) predicted CIE erythem ally w eighted UV dose (kJ/m 2)

Composite A Composite B

490 180

170
470

sod 160 sod

450
flexural strength (MPa)

flexural strength (Mpa)

jok jok
150
430 lin lin
isp 140 isp
410 the the
130
ela ela
390 iza
iza 120

370
110

350 100
0 200 400 600 800 1000 0 200 400 600 800 1000
predicted CIE erythem ally w eighted UV dose (kJ/m 2) predicted CIE erythem ally w eighted UV dose (kJ/m 2)

Figure 4. Property changes observed in the exposed samples over the first three exposure
periods as functions of predicted accumulation of UV dose.

The same features as revealed by the medians studied above may be detected in the graphs
of Fig. 4 as well. On the basis of the graphs, UV stress could be expected to explain most of
the yellowing. For the other property changes, UV is clearly not the only factor attributing
to the changes.

3.3 Spearman rank correlation analysis

To proceed towards the estimation of the most important agents, Spearman rank correlation
analysis is next performed as described in section 2.5. The correlation coefficients are
shown in Table 6. The statistically very significant (α = 0.05) coefficients are surrounded
with a thick-line rectangular and the statistically significant (α = 0.01) coefficients with a
thin-line rectangular.

Table 6. Spearman rank correlation coefficients computed for the ranks of the exposures
and the relative property changes; a) for GFR polyester (Composite A) and b) for GFR
vinyl ester (Composite B).
a) UV T P RH O3
Yellowness index 0.905 0.566 0.474 0.087 0.465
Gloss -0.430 -0.074 -0.044 0.299 0.003
Hardness 0.282 -0.175 -0.340 -0.399 0.049
 Flexural strength -0.443 -0.679 -0.488 -0.377 -0.601
b) UV T P RH O3
Yellowness index 0.934 0.606 0.557 0.121 0.491
Gloss -0.138 -0.021 0.144 0.116 -0.095
Hardness 0.004 -0.222 -0.060 -0.384 -0.465
Flexural strength -0.195 -0.274 -0.203 0.060 -0.097
According to the Spearman rank correlation test, UV stress indeed is the major cause for
yellowing of the both composites. However, stress caused by heat, rain and ozone
concentration of the air may also explain some of the observed color change. Changes in
gloss seem to be explained by the UV stress for GFR polyester. Decrease in flexural
strength of GFR polyester might be mainly attributed to the heat stress, and in some extent
to ozone, rain and UV. Moisture seems to be significant only in respect of changes in
hardness. Tropospheric ozone partly explains the changes in hardness of GFR vinyl ester
but not GFR polyester. For changes in gloss and flexural strength of GFR vinyl ester, no
particular agents can be found through this analysis.

3.4 Questions left open

A question arises as to whether the computed statistical figures as such adequately well
describe the prevailing weathering conditions or not. In the current figures, temporal
variations are not taken into account in any way. Yet especially for composite materials,
these kinds of features in the ageing environment may play an important role in the ageing
behaviour of the material.
The performed Spearman rank correlation analysis fails to catch the most effective stress
factors for changes in hardness and flexural strength of GFR vinyl ester. This might
indicate that these changes could be attributed to a synergistic effect of two or several
agents, which cannot be detected by the analysis method used.

4. CONCLUSIONS
Polyester and vinyl ester composites differ from each other in their response to
environmental exposure. Only changes in gloss seem to be similar for the both types of
composites. An initial increase, followed by a decrease in gloss, is observed. Polyester
composite shows more severe yellowing than vinyl ester. Polyester composite also shows
decrease both in surface hardness and flexural strength. Hardness and flexural strength of
vinyl ester composite increases first but then starts to decrease. The results are mainly in
line with the previous findings (e.g., [3,4]). Polyester resin matrix appears to be slightly
more susceptible to all significant environmental stress factors (UV radiation, precipitation,
humidity, temperature and ozone concentration) than vinyl ester. This is probably due to
the fact that the lack of non-terminal ester groups makes vinyl ester less vulnerable to
hydrolytic degradation [5]. Furthermore, vinyl ester has a higher glass transition
temperature Tg than polyester [5] and hence exhibits a higher resistance to certain (Norrish
II) types of photochemical processes [6].
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support from the Finnish Meteorological
Institute and the Finnish Funding Agency for Technology and Innovation (Tekes). The
personnel by the observatories within the UVEMA network is also acknowledged.

REFERENCES
1- Heikkilä, A., Tanskanen, A., Kärhä, P., and Hanhi, K., “Adjusting timing of
weathering test to account for seasonal variations in UV exposure”, Polymer
Degradation and Stability, 2007;92:675-683.
2- Cho, J.-D., Kim, H.-K., Kim, Y.-S., and Hong, J.-W., “Dual curing of cationic UV-
curable clear and pigmented coating systems photosensitized by thioxanthone and
anthracene”, Polymer Testing, 2003;22:633-645.
3- Correia, J.R., Cabral-Fonseca, S., Branco, F.A., Ferreira, J.G., Eusébio, M.I., and
Rodrigues, M.P., “Durability of pultruded glass-fibre-reinforced polyester profiles
for stuctrural applications”, Mech Compos Mater, 2006;42:325-338.
4- Signor, A.W., VanLandingham, M.R., and Chin, J.W., “Effects of ultraviolet
radiation exposure on vinyl ester resins: characterization of chemical, physical and
mechanical damage”, Polymer Degradation and Stability, 2003;79:359-368.
5- Boinard, E., Pethrick, R.A., Dalzel-Job, J., and MacFarlane, C.J., “Influence of resin
chemistry on water uptake and environmental ageing in glass fibre reinforced
composites-polyester and vinyl ester laminates”, Journal of Material Science,
2000;35:1931-1937.
6- Rabek, J.F., Polymer Photodegradation Mechanisms and experimental methods. 1st
edition, Chapman&Hall, University Press, Cambridge, 1995.